Uncertainty Estimation on AOKVQA
79.8AUCVisual Semantic Entropy
Evaluation Results
| Method | Links | |
|---|---|---|
| Visual Semantic EntropyCategory=Entropy, Venue=Ours, Backbone=Qwen3-VL2026.06 | 79.8 | |
| Visual Semantic EntropyCategory=Entropy, Venue=Ours, Backbone=Intern3.5-VL2026.06 | 79.2 | |
| Visual Semantic EntropyCategory=Entropy, Venue=Ours, Backbone=Qwen2.5-VL2026.06 | 78.3 | |
| VL-UCategory=Entropy, Venue=arxiv, Backbone=Intern3.5-VL2026.06 | 78.1 | |
| Visual Semantic EntropyCategory=Entropy, Venue=Ours, Backbone=Gemma32026.06 | 77.8 | |
| SECategory=Entropy, Venue=ICLR’23, Backbone=Intern3.5-VL2026.06 | 77.5 | |
| KLECategory=Entropy, Venue=NIPS’25, Backbone=Qwen2.5-VL2026.06 | 77.4 | |
| KLECategory=Entropy, Venue=NIPS’25, Backbone=Qwen3-VL2026.06 | 77.2 | |
| KLECategory=Entropy, Venue=NIPS’25, Backbone=Intern3.5-VL2026.06 | 76.4 | |
| VL-UCategory=Entropy, Venue=arxiv, Backbone=Qwen2.5-VL2026.06 | 75.6 | |
| SNNECategory=Entropy, Venue=ACL’25, Backbone=Qwen3-VL2026.06 | 74.7 | |
| SECategory=Entropy, Venue=ICLR’23, Backbone=Qwen3-VL2026.06 | 74.6 | |
| SNNECategory=Entropy, Venue=ACL’25, Backbone=Intern3.5-VL2026.06 | 74.5 | |
| SNNECategory=Entropy, Venue=ACL’25, Backbone=Qwen2.5-VL2026.06 | 74.4 | |
| SCG-PrCategory=Consistency, Venue=EMNLP’23, Backbone=Qwen2.5-VL2026.06 | 73.4 | |
| C&UCategory=Consistency, Venue=CVPR’24, Backbone=Intern3.5-VL2026.06 | 73.4 | |
| SECategory=Entropy, Venue=ICLR’23, Backbone=Gemma32026.06 | 73.2 | |
| C&UCategory=Consistency, Venue=CVPR’24, Backbone=Qwen2.5-VL2026.06 | 73.1 | |
| SCG-PrCategory=Consistency, Venue=EMNLP’23, Backbone=Intern3.5-VL2026.06 | 73 | |
| Visual Semantic EntropyCategory=Entropy, Venue=Ours, Backbone=LLaVA-NeXT2026.06 | 72.4 | |
| KLECategory=Entropy, Venue=NIPS’25, Backbone=Gemma32026.06 | 72.1 | |
| C&UCategory=Consistency, Venue=CVPR’24, Backbone=Qwen3-VL2026.06 | 72 | |
| VL-UCategory=Entropy, Venue=arxiv, Backbone=Gemma32026.06 | 71.5 | |
| C&UCategory=Consistency, Venue=CVPR’24, Backbone=Gemma32026.06 | 71.2 | |
| SCG-PrCategory=Consistency, Venue=EMNLP’23, Backbone=Qwen3-VL2026.06 | 70.8 | |
| VL-UCategory=Entropy, Venue=arxiv, Backbone=Qwen3-VL2026.06 | 70.6 | |
| SECategory=Entropy, Venue=ICLR’23, Backbone=Qwen2.5-VL2026.06 | 70.2 | |
| SCG-PrCategory=Consistency, Venue=EMNLP’23, Backbone=Gemma32026.06 | 70.1 | |
| Verb-UCategory=Verbalized, Venue=NAACL’24, Backbone=Qwen3-VL2026.06 | 70 | |
| SNNECategory=Entropy, Venue=ACL’25, Backbone=Gemma32026.06 | 70 | |
| Verb-UCategory=Verbalized, Venue=NAACL’24, Backbone=Gemma32026.06 | 65.4 | |
| SNNECategory=Entropy, Venue=ACL’25, Backbone=LLaVA-NeXT2026.06 | 65.1 | |
| MaxEntCategory=Logit, Venue=EMNLP’24, Backbone=LLaVA-NeXT2026.06 | 63.8 | |
| Verb-UCategory=Verbalized, Venue=NAACL’24, Backbone=Qwen2.5-VL2026.06 | 63.5 | |
| AvgEntCategory=Logit, Venue=EMNLP’24, Backbone=LLaVA-NeXT2026.06 | 63.5 | |
| AvgProbCategory=Logit, Venue=EMNLP’24, Backbone=LLaVA-NeXT2026.06 | 63.4 | |
| MaxEntCategory=Logit, Venue=EMNLP’24, Backbone=Intern3.5-VL2026.06 | 63.2 | |
| Verb-UCategory=Verbalized, Venue=NAACL’24, Backbone=Intern3.5-VL2026.06 | 63 | |
| AvgEntCategory=Logit, Venue=EMNLP’24, Backbone=Qwen2.5-VL2026.06 | 62.5 | |
| SECategory=Entropy, Venue=ICLR’23, Backbone=LLaVA-NeXT2026.06 | 62.2 | |
| SCG-NLICategory=Consistency, Venue=EMNLP’23, Backbone=Intern3.5-VL2026.06 | 61.8 | |
| VL-UCategory=Entropy, Venue=arxiv, Backbone=LLaVA-NeXT2026.06 | 61.6 | |
| AvgEntCategory=Logit, Venue=EMNLP’24, Backbone=Intern3.5-VL2026.06 | 61.3 | |
| AvgProbCategory=Logit, Venue=EMNLP’24, Backbone=Qwen2.5-VL2026.06 | 61.2 | |
| SCG-PrCategory=Consistency, Venue=EMNLP’23, Backbone=LLaVA-NeXT2026.06 | 60.8 | |
| C&UCategory=Consistency, Venue=CVPR’24, Backbone=LLaVA-NeXT2026.06 | 60.2 | |
| MaxProbCategory=Logit, Venue=EMNLP’24, Backbone=Qwen2.5-VL2026.06 | 60 | |
| MaxEntCategory=Logit, Venue=EMNLP’24, Backbone=Qwen2.5-VL2026.06 | 59.6 | |
| AvgProbCategory=Logit, Venue=EMNLP’24, Backbone=Intern3.5-VL2026.06 | 59.3 | |
| SCG-NLICategory=Consistency, Venue=EMNLP’23, Backbone=Qwen3-VL2026.06 | 59.1 | |
| AvgEntCategory=Logit, Venue=EMNLP’24, Backbone=Qwen3-VL2026.06 | 58.1 | |
| Verb-UCategory=Verbalized, Venue=NAACL’24, Backbone=LLaVA-NeXT2026.06 | 58 | |
| SCG-NLICategory=Consistency, Venue=EMNLP’23, Backbone=LLaVA-NeXT2026.06 | 57.9 | |
| MaxEntCategory=Logit, Venue=EMNLP’24, Backbone=Qwen3-VL2026.06 | 57.8 | |
| SCG-NLICategory=Consistency, Venue=EMNLP’23, Backbone=Gemma32026.06 | 57.6 | |
| SCG-NLICategory=Consistency, Venue=EMNLP’23, Backbone=Qwen2.5-VL2026.06 | 57.5 | |
| MaxProbCategory=Logit, Venue=EMNLP’24, Backbone=Gemma32026.06 | 56.1 | |
| AvgProbCategory=Logit, Venue=EMNLP’24, Backbone=Qwen3-VL2026.06 | 55.6 | |
| MaxProbCategory=Logit, Venue=EMNLP’24, Backbone=Qwen3-VL2026.06 | 54.1 | |
| KLECategory=Entropy, Venue=NIPS’25, Backbone=LLaVA-NeXT2026.06 | 54.1 | |
| AvgEntCategory=Logit, Venue=EMNLP’24, Backbone=Gemma32026.06 | 53 | |
| MaxEntCategory=Logit, Venue=EMNLP’24, Backbone=Gemma32026.06 | 52.9 | |
| MaxProbCategory=Logit, Venue=EMNLP’24, Backbone=LLaVA-NeXT2026.06 | 52.2 | |
| MaxProbCategory=Logit, Venue=EMNLP’24, Backbone=Intern3.5-VL2026.06 | 51.2 | |
| AvgProbCategory=Logit, Venue=EMNLP’24, Backbone=Gemma32026.06 | 50.8 |